get_device_memory_objects(): dispatch on cudf.core.frame.Frame

[madsbk]

get_device_memory_objects() now dispatch on cudf.core.frame.Frame instead of cudf.core.dataframe.DataFrame only.

This includes the CI fix, closes #659

[madsbk]

Current CI fail relates to rapidsai/cudf#8570

[quasiben]

@charlesbluca can you take a look at this ?

[charlesbluca]

Sure! @madsbk do you have a link to the test failure? I am only seeing passed and pending checks on my end

[madsbk]

rerun tests

[quasiben]

It looks like things are failing in distributed tests:

04:08:33 file /opt/conda/envs/rapids/lib/python3.7/site-packages/distributed/tests/test_nanny.py, line 444
04:08:33   @pytest.mark.asyncio
04:08:33   async def test_nanny_closes_cleanly(cleanup):
04:08:33       async with Scheduler() as s:
04:08:33           n = await Nanny(s.address)
04:08:33           assert n.process.pid
04:08:33           proc = n.process.process
04:08:33           await n.close()
04:08:33           assert not n.process
04:08:33           assert not proc.is_alive()
04:08:33           assert proc.exitcode == 0
04:08:33 E       fixture 'cleanup' not found
04:08:33 >       available fixtures: anyio_backend, anyio_backend_name, anyio_backend_options, benchmark, benchmark_weave, cache, capfd, capfdbinary, caplog, capsys, capsysbinary, cov, doctest_namespace, event_loop, monkeypatch, no_cover, pytestconfig, record_property, record_testsuite_property, record_xml_attribute, recwarn, testrun_uid, tmp_path, tmp_path_factory, tmpdir, tmpdir_factory, unused_tcp_port, unused_tcp_port_factory, worker_id
04:08:33 >       use 'pytest --fixtures [testpath]' for help on them.

These test pass locally so. I'll rerun now

[quasiben]

rerun tests

[codecov-commenter]

Codecov Report

❗ No coverage uploaded for pull request base (branch-21.08@f3220a4). Click here to learn what that means.
The diff coverage is n/a.

@@               Coverage Diff               @@
##             branch-21.08     #658   +/-   ##
===============================================
  Coverage                ?   61.75%           
===============================================
  Files                   ?       21           
  Lines                   ?     2604           
  Branches                ?        0           
===============================================
  Hits                    ?     1608           
  Misses                  ?      996           
  Partials                ?        0           

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update f3220a4...fbff1ee. Read the comment docs.

[quasiben]

Thanks @madsbk

[quasiben]

@gpucibot merge

[ajschmidt8]

approved with a suggestion 👍

[ajschmidt8]

Suggested change

	py.test --cache-clear -vs distributed/distributed/protocol/tests/test_cupy.py
	py.test --cache-clear -vs distributed/distributed/protocol/tests/test_numba.py
	py.test --cache-clear -vs distributed/distributed/protocol/tests/test_rmm.py
	py.test --cache-clear -vs distributed/distributed/protocol/tests/test_collection_cuda.py
	py.test --cache-clear -vs distributed/distributed/tests/test_nanny.py
	py.test --cache-clear -vs distributed/distributed/diagnostics/tests/test_nvml.py
	pytest --cache-clear -vs distributed/distributed/protocol/tests/test_cupy.py
	pytest --cache-clear -vs distributed/distributed/protocol/tests/test_numba.py
	pytest --cache-clear -vs distributed/distributed/protocol/tests/test_rmm.py
	pytest --cache-clear -vs distributed/distributed/protocol/tests/test_collection_cuda.py
	pytest --cache-clear -vs distributed/distributed/tests/test_nanny.py
	pytest --cache-clear -vs distributed/distributed/diagnostics/tests/test_nvml.py

While we're touching these lines, can we update these commands to pytest instead of py.test? It seems that py.test may be deprecated according to the SO discussion below.

https://stackoverflow.com/q/39495429

[ajschmidt8]

whoops. forgot that @gpucibot merge would merge after my approval. this can be addressed in a future PR I guess.